© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Under Peer Review · Sparse Autoencoders for Pythia-70M-Deduped
    5. Pythia-70M-Deduped
    7. Residuals
    9. 2-RES-SM
    11. 10897
    Prev
    Next
    INDEX
    Explanations

    the presence of significant acronyms or abbreviations

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    ctigges/pythia-70m-deduped__res-sm_processed/2-res-sm
    Prompts (Dashboard)
    32,768 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    32,768
    Data Type
    torch.float32
    Hook Name
    blocks.2.hook_resid_post
    Hook Layer
    2
    Architecture
    standard
    Context Size
    128
    Dataset
    EleutherAI/the_pile_deduplicated
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    369
    +0.30
    1.7%
    156
    +0.19
    1.1%
    410
    +0.14
    0.8%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    369
    +0.30
    0.04
    410
    +0.19
    0.05
    201
    +0.14
    0.03
    Negative Logits
    ception
    -1.98
    cludes
    -1.89
    cession
    -1.86
    lection
    -1.80
    ductive
    -1.75
    letion
    -1.74
    iquity
    -1.73
    cip
    -1.69
    ploy
    -1.68
    cess
    -1.67
    POSITIVE LOGITS
    ĻĤ
    2.00
    Ĵ
    1.98
    ĨĴ
    1.93
    ĭ
    1.65
    ŀ
    1.60
    Ĥ¬
    1.58
    Į
    1.57
    inetics
    1.57
    ±
    1.56
    !
    1.55
    Activations Density 0.104%

    No Known Activations